Intro

Notebook exploring some aspects of the supply & demand survey data.

Last updated: 2022-06-01

Exploration

Table: responses and animals

This is just to give some context to all subsequent figures. Covers the number of survey responses, current animals, and animals left in the last 5 years.

# Current Animals
# Left Last 5 Years
Study Community Responses Total Dogs Cats Total Dogs Cats
Cabot, Arkansas 864 1793 1108 685 1488 926 562
Dallas 993 1687 1089 598 1429 872 557
Fresno, CA 875 1875 1145 730 1624 907 717
Metro Detroit 1113 1555 878 677 1658 840 818
New Hampshire 871 1349 664 685 872 446 426
Palm Valley 629 1418 934 484 1070 701 369
Washington DC 973 1354 799 555 1353 738 615

Table: frequency of having and feeding animals

The format is: raw number (percentage of responses in that study community).

The groups that appear later show on some of these as well (DC, Detroit, and NH on one hand, Fresno and PV the other, Dallas and Cabot in between).

Study Community Responses Have Dog(s) Have Cat(s) Have Either Have Both Feed Cats
Cabot, Arkansas 864 581 (67.2%) 353 (40.9%) 674 (78%) 260 (30.1%) 264 (30.6%)
Dallas 993 647 (65.2%) 378 (38.1%) 736 (74.1%) 289 (29.1%) 315 (31.7%)
Fresno, CA 875 623 (71.2%) 354 (40.5%) 722 (82.5%) 255 (29.1%) 287 (32.8%)
Metro Detroit 1113 586 (52.7%) 393 (35.3%) 734 (65.9%) 245 (22%) 295 (26.5%)
New Hampshire 871 448 (51.4%) 392 (45%) 628 (72.1%) 212 (24.3%) 130 (14.9%)
Palm Valley 629 482 (76.6%) 240 (38.2%) 531 (84.4%) 191 (30.4%) 295 (46.9%)
Washington DC 973 539 (55.4%) 351 (36.1%) 663 (68.1%) 227 (23.3%) 242 (24.9%)

Animals leaving home: per response, by county

The following figure shows the number of animals/dogs/cats (see different tabs) left home in the last 5 years divided by the number of responses per county.

Minor preprocessing notes:

  1. Responses with more than 50 animals leaving home in the last 5 years were excluded (I just chose an arbitrary threshold, which removed 5 responses; if the threshold was 30 it would exclude 4 more responses when considering only dogs, for example).

  2. The figures only shows data for the 17 (of 119) counties that had more than 100 survey responses. If the threshold was 50, we would have 30 counties.

Things that stood out: California and Texas counties star, with Hidalgo (TX), Tulare (CA) and Fresno (CA) at the top across species; Tulare high in both species, Hidalgo leading the dogs, Fresno primarily cats. Cabot high up for dogs as well.

All Species

Dogs

Cats

Animals leaving home: per current animal, by county/community

The following figure shows the number of animals/dogs/cats (see different tabs) left home in the last 5 years divided by 5 (to get a rough estimate of # animals left in a year, assuming constant rate) over the number of current animals.

Same preprocessing notes as above apply here.

I also added a few extra tabs showing this by study community.

County, All

County, Dogs

County, Cats

Community, All

Community, Dogs

Community, Cats

Age: by community

Looking at age differences between study communities for currently-owned animals.

Dogs: DC, Detroit, and NH seem slightly ‘older’ compared to Palm Valley (highest % of <1, lowest of all categories of 6 years and older). Cabot, Dallas, and Fresno are in between.

Cats: Palm Valley is again the ‘youngest’, NH the oldest, and the others are roughly (though not as clearly) grouped as before - Fresno (and to som extent Dallas and Cabot) are on the young side, DC and Detroit a bit older (higher % of 11+, lower percentages of kittens).

All species

Dogs

Cats

Age when joined household: by community

Like the previous section, but looking at the age the animal was joined the household rather than the current age.

Dogs:

  1. Some variation in the % of puppies, but I’m not sure it’s significant - Palm Valley and Fresno have 72-73% of <1 puppies as opposed to 64-65% in Cabot and Dallas, the rest are in between with 67-69%.
  2. Fairly similar across the 6-10, 11-15, and 16+ groups.

Cats: highest <1 kittens % in Fresno followed by Palm Valley, but other than that, fairly small differences.

All species

Dogs

Cats

Spay/Neuter status: by community/county

% of animals spayed or neutered by study community (now added by county as well).

For both species, Palm Valley is lowest on % altered (by what seems like a significant margin), followed by Cabot and Fresno. For Dogs, Dallas is more similar to DC, Detroit and NH, whereas for Cats there is a ~5 percentage point gap from all three (and 10 percentage points higher than Cabot and Fresno).

Community, All

Community, Dogs

Community, Cats

County, All

County, Dogs

County, Cats

Ways animals left home: by community

The % of animals who left home in a particular way within each study community. This time excludinig >50 animals left home responses but including all counties.

Some notes:

  1. Significant variation in the % put to sleep - from a low of 8-12% in PVAS, Fresno, Cabot (where most animals left generally) to highs 22-32% in DC, Detroit, and NH. These numbers are for all species, but the gap exists when looking at dogs and cats separately as well.

  2. While a smaller margin, these two groups also differ in the % of animals given away in the other direction - 23-24% in Fresno, Cabot, PV and 15-19% in DC, Detroit, DC.

  3. NH stands out with lowest running away percentage (8%) compared to all other communities (16-24%). Even more intense when looking only at dogs - 1.5% in NH, 14-20% in all the others.

All Species

Dogs

Cats

Ways animals left home: by income

This is the same as the previous section, but with income brackets being the grouping variable.

The % differences aren’t very pronounced in all cases, but it seems like the % of animals given away is higher for lower income brackets and decreases with income (not monotonically, roughly), whereas the % of animals put to sleep (and somewhat also those dying) shows the opposite trend. Which makes sense to me. % running away is highest for <15k but 2nd highest for 150k+. This trend is true for cats and dogs.

All animals

Dogs

Cats

Animals given away: by reason, by community

I wanted to look at the reasons cited for giving away animals when that happened. 1585 animals had a non-null reason listed, in all of which the owner’s survey also indicated an animal given away (no mismatches here). In addition, out of the 5935 NA reason, 1200 animals (associated with 543 respondents) were listed without reason despite owner indicating an animal was given away in their response.

So out of 1416 people who indicated giving away 2785 animals in total, 873 people listed a reason for 1585 animals, i.e. reasons are available for 57% of given away animals and given by 62% of people who gave away animals. I’m not sure if you think it’s a good enough percentage, but I went ahead anyway.

Some notes (all species):

  1. NH stands out for housing (‘lease restrictions’), accounting for 22% of animals given away (24% of dogs), followed by Cabot and Dallas (16%).

  2. Detroit and DC stand out with behavior being the leading cause (24 and 21%).

  3. Cabot, Fresno, and PVAS have the highest percentages of animals given away because they were temporarily held (25-28%, vs. 18-20% in the other communities).

When looking only at dogs, PVAS temporary % is much lower, cost (23%) and behavior (20%) are the main reasons. Likewise, for cats only, the temporary % is higher in all three (30-35%) while remaining lower (14-20%) in the other communities.

All species

Dogs

Cats

Animals given away: by destination, by community

This is similar to the above section, but looking at who the animal was given to rather than the reason listed for giving the animal away. Whenever a reason was listed a destination was also listed, so the percentages of completion are the same as the above (57% of animals given away and 62% of people who gave away animals have completed these).

DC and Detroit stand out as having 51% and 54% of animals given to friends or relatives as opposed to 63% in Dallas and 69-74% in all others. The difference is animals given to rescues or shelters.

All species

Dogs

Cats

Way of getting animals: by community

I also took a look at how people got their currnet pets. I grouped the survey responses as follows:

  • Breeder, craigslist, facebook (if mentioned in ‘other’), purchase pet store - Purchased.
  • Found/stray, ‘rescued’ mentioned in ‘other’ - Found / Rescued.
  • Bred from own, ‘born in’ mentioned in ‘other’ - Previous Pet.
  • Shelter and Rescue grouped together.
  • Friend/family, Other remained their own categories.

Obviously, this is just one way to group them, I can see a case for others (e.g. craigslist/facebook separated from breeders or pet stores because of their peer-to-peer nature).

All species

When looking at all species, I found it interesting that the same groups (DC, NH, Detroit on one hand, Fresno and PVAS on the other, Dallas here very similar to Detroit and Cabot more like Fresno and PV) emerge with higher rates of obtaining pets from shelters, as well as buying pets, and lower rates of finding/rescuing and obtaining them via previous pets.

Also, everywhere but PV, more people adopted animals from shelters/rescues than purchased them, and in Dallas, DC and NH this was the leading source of pets (in Detroit it was tied with friends/family).

Dogs

For dogs, the difference of receiving from friends and family is pronounced (leading way in PV, Fresno, Cabot, much lower in NDD).

Cats

For Cats, finding and rescuing loose animals is much higher than in dogs, as expected, and much more common in PV, Fresno and Cabot than the rest, while adopting from shelters and rescues is much less common compared to the rest. Purchasing is much less common across the board.

Way of getting animals: by income

All species

A few things come up:

  1. The % of animals obtained from friends and family and found / rescued monotonically decreases as income bracket increases. From previous pet has more of a break after the first 2 brackets and remains the same.

  2. The % of animals purchased is monotonically increasing with income, as expected, but so is the % of animals obtained from a shelter/rescue.

Dogs

This is fairly similar to the overall figure, with the trends for friend/family decrease and purchase + shelter/rescue increase even more pronounced in light of the lower % for other categories.

Cats

Here the clearer trend is the decrease in the % of cats found or rescued as strays. Also, the two lower brackets have higher % of ‘previous pets’ while the two highest ones show a jump in purchasing percentages.

Modeling

Predicting an animal running/given away

Reference categories: <15k income, white, own, attached multifamily with outdoor access, Cabot.

  Pet Running/Given Away (Binary)
Predictors Estimates CI p
(Intercept) 0.41 0.36 – 0.46 <0.001
income_15-35k -0.03 -0.07 – 0.02 0.265
income_35-50k -0.05 -0.10 – -0.00 0.035
income_50k-75k -0.10 -0.14 – -0.05 <0.001
income_75k-100k -0.04 -0.09 – 0.01 0.097
income_100k-150k -0.02 -0.07 – 0.03 0.395
income [150k+] -0.03 -0.08 – 0.03 0.334
race [other] 0.05 0.00 – 0.10 0.048
race [latino] 0.11 0.07 – 0.16 <0.001
race [black] 0.11 0.06 – 0.15 <0.001
housing [attached
multifamily without
outdoor access] 		-0.06 -0.11 – -0.01 0.017
housing [detached single
family] 		-0.09 -0.12 – -0.06 <0.001
housing [other] -0.04 -0.10 – 0.03 0.305
study comm [Dallas] -0.01 -0.06 – 0.03 0.634
study comm [Fresno, CA] -0.05 -0.10 – -0.01 0.028
study comm [Metro
Detroit] 		-0.08 -0.13 – -0.04 <0.001
study comm [New
Hampshire] 		-0.17 -0.22 – -0.12 <0.001
study comm [Palm Valley] -0.03 -0.08 – 0.03 0.349
study comm [Washington
DC] 		-0.11 -0.16 – -0.06 <0.001
Observations 4836
R2 / R2 adjusted 0.046 / 0.042

Predicting animal ownership

On the left - DV is a binary for having a cat or a dog. Right - having a cat or dog or feeding >=1 neighborhood cat.

Reference categories: <15k income, white, own, attached multifamily with outdoor access, Cabot.

  Having Pets (Binary) Having/Feeding Pets (Binary)
Predictors Estimates CI p Estimates CI p
(Intercept) 0.82 0.77 – 0.87 <0.001 0.86 0.81 – 0.90 <0.001
income_15-35k 0.03 -0.01 – 0.07 0.141 0.01 -0.02 – 0.05 0.484
income_35-50k 0.05 0.01 – 0.10 0.019 0.03 -0.01 – 0.07 0.165
income_50k-75k 0.04 -0.00 – 0.08 0.059 0.01 -0.03 – 0.05 0.571
income_75k-100k 0.07 0.02 – 0.11 0.003 0.04 -0.00 – 0.09 0.057
income_100k-150k 0.09 0.04 – 0.13 <0.001 0.06 0.02 – 0.11 0.004
income [150k+] 0.09 0.04 – 0.14 <0.001 0.06 0.02 – 0.11 0.009
race [other] -0.09 -0.13 – -0.05 <0.001 -0.06 -0.10 – -0.02 0.002
race [latino] 0.02 -0.01 – 0.06 0.219 0.04 0.00 – 0.08 0.030
race [black] -0.13 -0.17 – -0.09 <0.001 -0.08 -0.12 – -0.05 <0.001
own rent [rent] -0.05 -0.08 – -0.03 <0.001 -0.02 -0.05 – 0.00 0.102
housing [attached
multifamily without
outdoor access] 		-0.11 -0.15 – -0.06 <0.001 -0.10 -0.14 – -0.06 <0.001
housing [detached single
family] 		-0.01 -0.04 – 0.01 0.320 -0.01 -0.04 – 0.01 0.366
housing [other] -0.15 -0.21 – -0.10 <0.001 -0.12 -0.17 – -0.06 <0.001
study comm [Dallas] -0.06 -0.11 – -0.02 0.002 -0.07 -0.11 – -0.03 0.001
study comm [Fresno, CA] 0.02 -0.02 – 0.06 0.366 0.01 -0.03 – 0.05 0.598
study comm [Metro
Detroit] 		-0.14 -0.18 – -0.10 <0.001 -0.14 -0.18 – -0.10 <0.001
study comm [New
Hampshire] 		-0.09 -0.13 – -0.05 <0.001 -0.11 -0.15 – -0.07 <0.001
study comm [Palm Valley] 0.01 -0.04 – 0.07 0.592 0.03 -0.02 – 0.08 0.286
study comm [Washington
DC] 		-0.13 -0.17 – -0.09 <0.001 -0.13 -0.17 – -0.09 <0.001
Observations 5930 5930
R2 / R2 adjusted 0.057 / 0.054 0.047 / 0.044